Type 1 Diabetic Akita Mouse Hearts Are Insulin Sensitive but Manifest Structurally Abnormal Mitochondria That Remain Coupled Despite Increased Uncoupling Protein 3

Male Akita mice (C57BL/6) and C57BL/6 controls were obtained from The Jackson Laboratories (Bar Harbor, ME) and housed at 22°C with free access to water and food with a light cycle of 12 h light and 12 h dark. Animals were studied in accordance with protocols approved by the Institutional Animal Care and Use Committee of the University of Utah. All studies were performed in random-fed animals.

Free fatty acid concentrations, triacylglycerol levels, and blood glucose levels were measured as previously described (4).

Contractile function and left ventricular chamber dimensions were measured as described previously (23).

Contractile function and substrate metabolism were measured in isolated working hearts perfused with Krebs-Henseleit buffer (KHB) containing 5 mmol/l glucose and 0.4 mmol/l palmitate in the presence or absence of 1 nmol/l insulin as previously described (3). For some mitochondrial respiration studies, hearts were perfused in the Langendorff mode with KHB containing 11 mmol/l glucose and 1 mmol/l palmitate for 30 min (5). The effect of a calcium-induced increase in workload on contractility was investigated in Langendorff-perfused hearts with KHB (5 mmol/l glucose and 0.4 mmol/l palmitate) (5). Hearts were perfused for 40 min, and calcium was raised from 2 to 4 mmol/l after 20 min.

Saponin-permeabilized fibers were prepared from freshly excised or Langendorff-preperfused hearts as described previously (5). Respiration and ATP synthesis were measured using 20 μmol/l palmitoyl carnitine, 10 mmol/l pyruvate, or 5 mmol/l glutamate as substrates, each combined with 2 mmol/l malate. Respiration was measured in the presence of substrate alone (V₀), after addition of 1 mmol/l ADP (V_ADP), and after addition of 1 μg/ml F₀F₁-ATPase inhibitor oligomycin (V_Oligo).

Proton leak measurements were performed in hearts that were preperfused for 20 min with 11 mmol/l glucose and 1 mmol/l palmitate. Mitochondria were prepared from whole hearts by differential centrifugation, and proton leak kinetics was measured in the presence of 20 μmol/l palmitoyl carnitine as previously described (6,24).

Hearts were freshly excised and immediately washed in ice-cold saline. Samples were collected from left ventricular myocardium and processed as described previously (6). Mitochondrial volume density and number were analyzed by stereology in a blinded fashion using the point-counting method (6,25).

Mitochondrial hydrogen peroxide (H₂O₂) generation was measured using succinate as a substrate as described previously (6). To block H₂O₂ production at complex I, 10 μmol/l rotenone was added to the reaction. Mitochondrial H₂O₂ production is widely accepted as a measure of mitochondrial superoxide production.

Tissue ROS levels were measured by the conversion of nonfluorescent 2′, 7′-dichlorofluorescein-diacetate (DCFDA) to the highly fluorescent 2′, 7′-dichlorofluorescein (DCF) in the presence of esterases and ROS (26) (for details, see the supplementary methods available in an online appendix at http://dx.doi.org/10.2337/db08-0079). Activity of mitochondrial aconitase was measured in isolated mitochondria as described previously (27).

Total RNA was extracted from hearts with Trizol reagent (Invitrogen, Carlsbad, CA), purified with the RNEasy kit (Qiagen, Valencia, CA), and reverse transcribed (6). Data were normalized by expressing them relative to the levels of the invariant transcript 16S RNA. Primer sequences and accession numbers are presented in supplementary Table 1.

Isolated mitochondria or whole-cell extracts were prepared as described previously and resolved by SDS-PAGE (5,6). For antibody conditions see supplementary methods in the online appendix. Protein concentrations were determined using the Micro BCA Protein Assay kit (Pierce, Rockford, IL), and equal protein loading was verified by Coomassie staining.

Data are presented as means ± SE. When comparing two groups, significance was determined using a Student's t test. Effects of calcium-induced increase in contractility were analyzed using a paired t test. Analysis of working heart perfusions ± insulin was performed using one-way ANOVA. Significance was assessed by Fisher's protected least significant difference test. For all analyses, the Statview 5.0.1 software package was used (SAS Institute, Cary, NC), and significant difference was accepted when P < 0.05.

Akita diabetic mice develop type 1 diabetes as early as 5–6 weeks of age and show normal survival until 6 months of age but then show a dramatic decrease in survival with almost no survivors by 12 months of age (19). To investigate the long-term effects of type 1 diabetes on the heart, we chose to investigate mice mainly at 24 weeks of age. Serum glucose levels were markedly increased in Akita mice at 24 weeks of age (138 ± 5 vs. 566 ± 21 mg/dl; P < 0.05). Hyperglycemia was accompanied by increased serum fatty acid levels (261 ± 43 vs. 537 ± 72 μmol/l; P < 0.05) and increased serum triacylglycerol levels (0.98 ± 0.11 vs. 2.40 ± 0.54 mmol/l; P < 0.05). In contrast to wild-type mice, Akita mice showed no increase in body weights as they aged (supplementary Table 2). Heart weights and tibia lengths were measured at 10, 24, and 54 weeks of age in the Akita mice (supplementary Table 2). Although heart-to-tibia length ratios were not different at 10 weeks, we observed reduced ratios at 24 and 54 weeks of age. Heart weight–to–body weight ratios were not reduced, indicating that the reduction in heart size parallels the general catabolic state of diabetic Akita mice.

Cardiac contractile function and chamber dimensions were examined in Akita diabetic mice by echocardiography at 20, 36, and 54 weeks of age. Fractional shortening and ejection fraction were not different at any age (Table 1), and there were no differences in chamber dimensions in diabetic and nondiabetic mice (supplementary Table 3). Cardiac output was increased at 20 weeks of age but was lower at 54 weeks of age in Akita mice, which was likely due to a moderate but nonsignificant reduction in heart rate. Because adaptations of the neurohumoral system can mask the presence of contractile dysfunction in vivo, we investigated contractile function in isolated working hearts in the absence and presence of insulin (Fig. 1A–C). In the absence of insulin, left ventricular developed pressure was reduced in 24-week-old Akita mice, and a trend toward increased cardiac output in Akita hearts led to no differences in cardiac power. Insulin had a positive inotropic effect in wild-type hearts, which was blunted in Akita mice. Heart rates (bpm) were not significantly changed from baseline by insulin (basal vs. insulin: wild type, 263 ± 8 vs. 281 ± 8; Akita, 290 ± 5 vs. 310 ± 9). To further evaluate myocardial contractile reserve, Langendorff-perfused hearts were examined under normal and increased calcium concentrations. Contractile parameters were similar in 15-week-old wild-type and Akita mice under baseline conditions (2 mmol/l calcium). Changing perfusion conditions to 4 mmol/l calcium significantly increased systolic pressure, developed pressure, and rate pressure product in wild-type and Akita mice by equivalent extents (supplementary Table 4). Thus, despite persistent hyperglycemia, Akita mice exhibited modest left ventricular dysfunction that was evident mainly in insulin-perfused working hearts. In contrast to ob/ob mice (5), short-term inotropic reserves were maintained.

Decreased cardiac efficiency is believed to be an important contributor to the development of contractile dysfunction in type 2 diabetic models and has been postulated to result from fatty acid–induced mitochondrial uncoupling (1). We therefore measured cardiac substrate utilization and oxygen consumption in isolated working hearts of 24-week-old Akita mice in the absence and presence of insulin. Palmitate oxidation rates were increased and glucose oxidation rates were decreased in the presence or absence of 1 nmol/l insulin in Akita hearts compared with the wild type (Fig. 1D and E). Despite increased fatty acid oxidation, myocardial Vo₂ and cardiac efficiency were not different between wild-type and Akita mice (Fig. 1G and H). Of interest, metabolic insulin responsiveness was retained in Akita hearts. Thus, insulin-suppressed palmitate oxidation increased glycolysis and glucose oxidation in both wild-type and Akita mice (Fig. 1D–F). Consistent with intact metabolic insulin signaling, insulin-stimulated Akt phosphorylation on Ser⁴⁷³ was maintained in hearts of 10- and 24-week-old mice (Fig. 2A and B). Thus, despite increased fatty acid utilization, Akita hearts maintained normal insulin sensitivity.

The absence of reduced cardiac efficiency was unexpected and revealed an important difference between the type 1 diabetic Akita mouse and type 2 diabetic mouse models. Because mitochondrial uncoupling may underlie impaired cardiac efficiency in type 2 diabetic hearts (1,5,6), we hypothesized that mitochondrial coupling would be normal in Akita hearts. Mitochondrial function was initially evaluated in fibers that were prepared from freshly isolated mouse hearts. In hearts from 24-week-old mice, maximal ADP-stimulated mitochondrial oxygen consumption (V_ADP) and ATP synthesis were unchanged using palmitoyl carnitine as a substrate (Fig. 3A and E), but V_ADP and ATP synthesis were significantly reduced in Akita mice using pyruvate (Fig. 3B and F) or glutamate (Fig. 3C and G) as substrates. Although a respiratory defect was evident, there was no evidence for mitochondrial uncoupling. Rates of oligomycin-insensitive respirations (V_oligo) and ATP-to-O ratios were unchanged with any substrate, indicating intact coupling of ATP synthesis to oxygen consumption. In 10-week-old mice, similar reductions in glutamate-supported respirations (V_ADP) were observed (Fig. 3D). ATP synthesis rates were proportionately reduced; thus, ATP-to-O ratios were not different (Fig. 3H).

We previously reported that mitochondrial uncoupling became evident in ob/ob and db/db mouse hearts after hearts were perfused with increased concentrations of fatty acids (5,6). Thus, palmitoyl carnitine respirations were determined in permeabilized fibers from hearts of 10-week-old Akita and control mice after perfusion with 11 mmol/l glucose and 1 mmol/l palmitate (Fig. 4A and B). Under these conditions, we observed no differences in V_ADP, V_oligo, ATP production rates, or ATP-to-O ratios. Finally, to provide more definitive evidence for the absence of mitochondrial uncoupling in Akita hearts, we isolated mitochondria from hearts of 24-week-old Akita mice after perfusion of hearts with 11 mmol/l glucose and 1 mmol/l palmitate and measured proton leak kinetics (Fig. 4C). Proton leak kinetics curves were entirely overlapping between mitochondria from Akita mice and nondiabetic controls, indicating the absence of mitochondrial uncoupling.

To understand the molecular basis for impaired mitochondrial respiratory capacity with glutamate and pyruvate, we investigated gene expression in Akita and wild-type hearts at 24 weeks of age (Fig. 5A). Expression of genes encoding for subunits of the electron transport chain was reduced for four of five genes examined in Akita mice. In contrast, two forms of peroxisome proliferator–activated γ coactivator-1 (PGC-1α and PGC-1β) and their downstream mediators (nuclear respiratory factor 1 [NRF1], mitochondrial transcription factor A [TFAm], and estogen-related receptor-α [ERRα]) were not differentially expressed. Genes encoding for peroxisome proliferator–activated receptor α (PPARα) and for enzymes and transporters of fatty acid utilization did not show a uniform trend toward increased expression except for malonyl CoA decarboxylase. Expression of GLUT1 and GLUT4 was reduced, and expression of pyruvate dehydrogenase kinase 4 (PDK4) was increased in Akita, consistent with impaired glucose oxidation in Akita. Interestingly, expression of UCP2 and UCP3 was significantly increased in Akita, and these changes were accompanied by increased expression of mitochondrial thioesterase 1 (MTE1). Consistent with increased mRNA levels, there was a twofold increase in UCP3 protein content (Fig. 5B). This observation is of particular interest because we detected no evidence of mitochondrial uncoupling in mitochondria from Akita mice.

Electron microscopy revealed markedly reduced cristae density in Akita but not in wild-type mice (Fig. 6A). Of the images evaluated, 75% showed lipid droplets in Akita mice, but no lipid droplets were observed in wild-type mice (arrows in Fig. 6A). Although mitochondrial volume density was increased in Akita mice, mitochondrial number was unchanged between the groups (Fig. 6B and C).

Thus far, we have shown that Akita mitochondria reveal defects in oxygen consumption in the absence of mitochondrial uncoupling (despite increased UCP3 content). We previously demonstrated that in type 2 diabetic db/db mice, fatty acid–induced mitochondrial uncoupling is mediated by activation of UCPs in the presence of increased ROS production and evidence of oxidative stress. These results suggested that increased ROS is required for activation of UCPs in diabetic hearts (6). We therefore hypothesized that the absence of mitochondrial uncoupling in Akita hearts despite increased UCP3 content was due to reduced ROS in Akita hearts. Consistent with this hypothesis, there was no increase in mitochondrial H₂O₂ production. Akita mice showed a 38–46% decrease in mitochondrial H₂O₂ generation compared with wild-type mice at 10 or 24 weeks of age (Fig. 7A and D). Addition of the complex I inhibitor rotenone abolished mitochondrial H₂O₂ production, suggesting that mitochondrial ROS production mainly arises from complex I in wild-type and Akita mice. Mitochondrial aconitase activity was not decreased in Akita mice at both ages (Fig. 7B and E), and oxidation of 2′-7′-dichlofluoroscein diacetate (DCFDA), a measure of tissue ROS levels, was unaltered in 24-week-old hearts (Fig. 7C). In addition, there was no induction of antioxidant defense pathways. Transcript levels of glutathione peroxidase 1, mitochondrial manganese superoxide dismutase (SOD2), and peroxiredoxin 3 (PRDX3) were either unchanged or even reduced in Akita (Fig. 5A). Similarly, protein levels of SOD2 and PRDX3 were unchanged between the groups (Fig. 7F and G).

Reduced cardiac efficiency is a well-described characteristic of type 2 diabetic hearts in animal models and humans (3,28). In this study, we addressed the following question: Do impaired cardiac efficiency and fatty acid–induced mitochondrial uncoupling contribute to cardiac contractile dysfunction in type 1 diabetic Akita mice? Similar to type 2 diabetic hearts, substrate metabolism shifts toward increased fatty acid metabolism, and mitochondrial oxidative capacity is impaired. However, in contrast to type 2 diabetic mouse models, Akita mice show relatively preserved contractile function under ambient conditions, no impairment in cardiac efficiency, and no evidence of mitochondrial uncoupling (normal V_oligo, ATP-to-O ratios, and proton leak). Perfusion of hearts with high fatty acid concentrations was not sufficient to induce mitochondrial uncoupling despite the presence of increased UCP3 levels. Moreover, the absence of oxidative stress distinguishes the Akita mouse from type 2 diabetic models. Thus, the molecular mechanisms for mitochondrial dysfunction importantly differ between insulin-deficient type 1 and insulin-resistant type 2 diabetic mice. We also show that increased UCP3 does not invariably lead to increased mitochondrial uncoupling in the heart, supporting the hypothesis that fatty acid–induced mitochondrial uncoupling in diabetic hearts may require a concomitant increase in ROS generation.

The absence of mitochondrial uncoupling in Akita hearts was associated with the absence of oxidative stress. Mitochondrial superoxide production, measured as H₂O₂ generation, was actually decreased, and the activity of mitochondrial aconitase and oxidation of DCFDA, which are independent measures of oxidative damage, were unaltered in Akita hearts. The association of mitochondrial uncoupling with increased ROS and lipid peroxide generation in type 2 diabetic db/db hearts (6) and the lack of uncoupling and the absence of oxidative stress in type 1 diabetic Akita hearts suggest a causal interdependence of ROS production and mitochondrial uncoupling in diabetic hearts. Superoxide increases proton conductance by UCP3 in isolated rat skeletal muscle mitochondria, and lipid peroxidation products, such as hydroxynonenal, increase proton leak in isolated mitochondria from multiple tissues, including the heart, via UCPs and adenine nucleotide translocator (7,24). Although fatty acids appear to be required for proton transport of UCP2 and UCP3 when reconstituted in liposomes, Echtay et al. (24) showed that hydroxynonenal-induced proton conductance does not require the presence of fatty acids (29). Thus, it appears that for fatty acid–induced mitochondrial uncoupling, as observed to occur in type 2 diabetic hearts, an increase in ROS may be required in addition to increased fatty acid availability and utilization.

The absence of oxidative stress or of any increase in ROS or H₂O₂ production in Akita hearts was unexpected. Inhibition of mitochondrial H₂O₂ production in the presence of rotenone suggested that complex I accounted for most of the ROS production in Akita and control hearts. This contrasts with similar studies in mitochondria isolated from db/db hearts in which additional mitochondrial complexes (likely complex III) also contributed to ROS generation (6). The reduction in ROS generation in Akita mice suggested that there might be a defect at the level of mitochondrial complex I. This was also supported by the observation that mitochondrial respirations in the presence of glutamate and pyruvate, which are complex I substrates, were also reduced, whereas this was not the case with palmitoyl carnitine that delivers reducing equivalents to complex I and complex II. Thus, a simple increase in substrate flux might not be sufficient to increase mitochondrial ROS generation in diabetic hearts in the absence of specific alterations in mitochondrial complex activities that amplify ROS generation.

Oxidative stress has been proposed to contribute to mitochondrial dysfunction in the hearts of type 1 diabetes mouse models (30,31). In the present study, we demonstrated impaired mitochondrial function and perturbed mitochondrial morphology in the absence of any evidence of oxidative stress. Interestingly, impaired mitochondrial morphology in the type 1 diabetic OVE26 mouse can be normalized by overexpression of manganese superoxide dismutase (MnSOD) or catalase in the heart, but MnSOD-deficient mice die from early dilated cardiomyopathy and have normal cardiac mitochondrial morphology (30–32). In other transgenic models, knockout of the mitochondrial transcription factor A leads to reduced mitochondrial DNA replication and transcription and enlarged mitochondria with abnormal cristae morphology, and combined deletion of the cardiac insulin and IGF1 receptor results in impaired oxidative phosphorylation (OXPHOS) gene transcription and decreased mitochondria that appeared to be less dense in electron micrographs (33–35). Thus, in some contexts, changes in the content of electron transport chain subunits can be associated with changes in mitochondrial morphology. Because insulin is a positive regulator of OXPHOS gene expression and because OXPHOS gene expression is decreased in the Akita mouse heart, insulin deficiency as opposed to oxidative stress might be an important contributor to mitochondrial dysfunction and altered morphology in the Akita mouse model (36,37). Defective insulin signaling has been reported to reduce expression levels of genes involved in β-oxidation (23), yet fatty acid oxidation genes were unchanged or increased in Akita mice. We believe that this reflects the impact of PPARα activation in the face of increased fatty acid availability.

An important observation in the present study is that increased content of UCP3 in the heart should not be taken to indicate increased mitochondrial uncoupling. We previously reported that increased mitochondrial uncoupling activity can occur in the absence of any changes in UCP3 content in hearts from ob/ob and db/db mice (5,6). It is important to discuss what might drive UCP3 levels in Akita mouse hearts. Besides detoxification of ROS, UCP3 has been postulated to play a role in the regulation of fatty acid metabolism. Conditions associated with increased fatty acid availability, such as fasting, high-fat feeding, and diabetes, result in increased UCP3 expression, likely due to increased PPARα signaling (12,38–40). In general, muscle UCP3 protein content is negatively related to fatty acid oxidative capacity, and induction of UCP3 is most pronounced in glycolytic muscle upon fasting and high-fat feeding (41,42). Thus, increased UCP3 expression may reflect an adaptive response to fatty acid overload (43). Based on the fatty acid cycling model, it has been proposed that export of fatty acid anions from the matrix into the intermembrane space via UCPs may help to lower intramitochondrial fatty acid levels if increased fatty acid availability exceeds mitochondrial oxidative capacity (43–45). Thereby, oxidation of intramitochondrial fatty acids into harmful lipid peroxides could be prevented. Expression of MTE1 often parallels that of UCP3, and this was also the case in the present study. MTE1 catalyzes conversion of acyl-CoAs back to fatty acid anions, which can then be translocated out of the matrix by UCPs (13,29). In the cytosol, they may be reesterified and oxidized in the mitochondria, stored as triglycerides, or used for other pathways. Nonesterified fatty acids can be generated within mitochondria and exported from the matrix, and these events are markedly increased in cardiac mitochondria from streptozotocin-induced type 1 diabetic rats (13). Despite increased basal fatty acid oxidation, Akita mice also revealed increased lipid droplets; thus, it is tempting to speculate that fatty acyl-CoAs would be converted to fatty acid anions by increased MTE1 activity, exported by UCP3, reesterified, and stored as triglycerides. Thus, UCPs would contribute to match fatty acid availability with oxidation to prevent intramitochondrial lipotoxic effects.

A number of important differences between Akita mice and previously evaluated models of type 2 diabetes were observed (3–6,8). First, there was a relative preservation of cardiac function and preserved inotropic responses to a calcium-induced increase in workload. This likely reflects the more profound mitochondrial dysfunction in the type 2 models and a greater degree of mitochondrial uncoupling. Preliminary studies do suggest, though, that cardiac dysfunction may develop in Akita mice after a longer-term hemodynamic challenge, such as after isoproterenol infusion (supplementary results and supplementary Table 5), but the mechanisms for this remains to be elucidated. Second, insulin sensitivity was preserved in Akita hearts in contrast to the insulin resistance reported in models of type 2 diabetes (2,3). Although insulin levels in diabetic Akita are 40% of insulin levels in nondiabetic controls (19), these ambient insulin concentrations might be sufficient to maintain mitochondrial function to a greater extent than models in which insulin signaling is severely impaired. Finally, myocardial Vo₂ was not increased despite increased oxidation rates of exogenous palmitate. Although this may partly reflect the absence of mitochondrial uncoupling, it is also possible that reduced oxidation of glucose and endogenous triglycerides coupled with mitochondrial dysfunction reduced myocardial Vo₂ in intact perfused hearts.

In conclusion, despite a significant increase in UCP3 content, insulin-deficient Akita hearts do not develop fatty acid–induced mitochondrial uncoupling, suggesting that underlying mechanisms for mitochondrial dysfunction may importantly differ between insulin-responsive type 1 versus insulin-resistant type 2 diabetic hearts. Increased UCP3 levels do not invariably lead to increased mitochondrial uncoupling in the heart, thereby supporting the hypothesis that fatty acid–induced mitochondrial uncoupling in diabetic hearts requires a concomitant increase in ROS or lipid peroxide generation.

H.B. has received a postdoctoral fellowship grant from the German Research Foundation. S.B. has received postdoctoral fellowships from the Juvenile Diabetes Research Foundation and the American Heart Association (AHA). V.G.Z. has received a postdoctoral fellowship from the AHA Western Affiliates. S.E.L. has received a VA Merit Award. E.D.A. is an established investigator of the AHA. This study was supported by National Institutes of Health Grants UO1-HL-70525, UO1-HL-087947 (AMDCC), and RO1-HL-070070.